Individual Submission Summary
Share...

Direct link:

Siblings’ Shared Interests on Twitter: A Case of Selecting a Sample of Dyads from Big Social Media Data

Fri, October 5, 10:45am to 12:15pm, Doubletree Hilton, Room: Redrock

Abstract

Siblings play significant roles in one another’s development (McHale et al., 2012). While most research focuses on youth, less is known about how siblings remain connected in young adulthood. The prevalence of social media, however may provide new opportunities (LeBouef & Dworkin, 2017). Twitter is a social media platform whose predominant audience is young adults (Pew Research Center, 2016). Prior research shows that Twitter can reflect couples’ connectedness (Garimella et al., 2014); we extended this to study siblings’ use of Twitter by analyzing the extent to which they share in following accounts, a proxy of their shared interests and a potential context for connectedness. As studies show that sister-sister dyads are closer than other sibling dyads (Kim et al., 2006), we examined dyad sex constellation differences in siblings’ overlapping following accounts to examine whether this pattern replicated on Twitter.

To identify sibling dyads, we used all geo-tagged tweets in the U.S. posted during December 16–31, 2015 (approximately 4.8 million; Chi et al., 2017). We chose this holiday period when families tend to come together and post their gatherings on social media. We extracted 4,722 tweets—those that contained one of the keywords, “sister(s)” “brother(s)” “sis” “bro(s)” “sissy” “sibling(s),” and in which a user mentioned another user (e.g., “Christmas with sister @user2name.”). Independent coders classified relationship of the two users within each tweet as sibling versus other based on the content (agreement rate=92%). This step identified 447 sibling dyads and labelled each sibling’s gender. We eliminated dyads in which at least one sibling’s Twitter account was protected, deleted, or inactive, resulting in 279 dyads for analyses.

To analyze siblings’ shared interests, we used Twitter’s API to extract users’ following accounts. The number of following accounts ranged from 1 to 25048 (median=421), and that of shared accounts ranged from 0 to 1196 (median=24). We calculated the overlap rate (number of shared following accounts/dyad’s total number of accounts) to represent siblings’ shared interests. The overlap rate distribution was non-normal, therefore we conducted permutation tests (Ernst, 2004) for cross-group comparisons (sister-sister vs. brother-brother vs. mixed-sex). This group effect was significant, p=.04; follow-ups show a higher overlap rate among brother-brother (M=.06, SD=.07) than mixed-sex dyads (M=.04, SD=.04), p=.01; sister-sister dyads did not differ from either (M=.05, SD=.07). Additional analyses address how shared interests are linked to the characteristics of siblings’ online interactions. Discussion address the use of social media data to study dyadic relationships.

Authors